Abstract: Reconstruction of fire history from stratigraphic charcoal records from lake sediment has focused on two levels of interpretation: the long-term trend in charcoal accumulation rates which may be aggregated into regional syntheses of biomass burning, and where possible, identified peaks of charcoal influx associated with events or episodes of fire within a region near the lake. Both approaches have relied on a series of assumptions and data manipulations. I present a new null-model approach that detects significant positive and negative deviations stratigraphic data of particle counts such as fossil pollen and charcoal. The method is especially relevant to contiguously sampled sediment charcoal data but can also be applied to any situation in which particles are enumerated in a dated stratigraphic section. The model simulates particle distribution with a Binomial process, and optionally includes effects of changing sedimentary rates on particle concentration. The null-model approach is applicable across a range of temporal scales and may be used to detect anomalies across temporal scales (e.g., in lake-sediment records, from decadal to millennial). The method requires a single parameter describing the overdispersion of particles, due to processes such as laboratory error and particle breakage. I apply the methods to charcoal stratigraphic data at two sites in east-central British Columbia and compare the results to existing methods of charcoal-peak analysis. At both sites, charcoal records show centennial-scale intervals between distinct peaks with little change over time until the last ca. 2000 years. After this, one site shows abrupt increases in peak size and frequency between 2000 to 650 yr BP, while at the other site the increase mainly occurs in the historic period. The climatic history, inferred from pollen and regional paleoclimate records, does not explain the late-Holocene increase in charcoal.
